The rapid digitization of academic credentialing has increased the need for scalable, secure, and verifiable blockchain infrastructures. Traditional consensus algorithms such as Proof of Work (PoW) and Practical Byzantine Fault Tolerance (PBFT) suffer from high energy consumption and limited scalability. This paper proposes EduDAG-PBFT, a novel hybrid consensus model that integrates a Directed Acyclic Graph (DAG) transaction structure with a lightweight PBFT mechanism optimized for large-scale educational systems. EduDAG-PBFT outperforms recent hybrids like SharDAG by achieving a 26 \(\times \) throughput improvement over traditional PBFT, while incorporating verifiable random functions (VRF) for enhanced committee security and zero-knowledge proofs (ZK-SNARKs) for privacy preservation. Through simulations involving 10,000 certificate transactions and a real-world pilot on 50 AWS nodes, the proposed model achieved 16,949 transactions per second (TPS) and completed processing in 0.59 s, outperforming traditional PBFT by more than 26 \(\times \) . The results demonstrate that EduDAG-PBFT significantly improves throughput, latency, scalability, and security for next-generation blockchain-enabled education platforms.